right cargo,

wrong address

Problem statement

In vivo gene therapy — correcting or replacing genes directly inside the patient's body — requires delivery vectors that simultaneously cross biological barriers (blood vessel walls, cell membranes, endosomal escape, nuclear entry), target the correct cell type with high specificity, carry sufficient genetic payload, and evade the immune system. No delivery vector achieves all four requirements. Viral vectors (AAVs) achieve good cell entry but trigger neutralizing antibodies that prevent repeat dosing, carry limited payload (<5 kb for AAV), and have tropism that is difficult to redirect to arbitrary cell types. Non-viral vectors (lipid nanoparticles) naturally accumulate in the liver after systemic administration, and delivery to non-liver tissues remains far less efficient.

Why this matters

Thousands of human diseases have identified genetic causes, but approved gene therapies number only in the low dozens — almost all for rare diseases affecting small patient populations. The IEEE grand-challenge authors put the cause plainly: "a lack of efficacious in vivo delivery tools, and the severe constraints on delivery and associated manufacturability have made most gene therapies limited in their capability and also prohibitively expensive" (Subramaniam et al. 2024, Grand Challenge 5, "Engineering Life – Engineering Genomes and Cells"). The current dominant approach is ex vivo therapy (removing cells, editing them outside the body, and reinfusing), which carries list prices in the millions of dollars per patient. In vivo delivery would eliminate the need for cell harvesting and reinfusion; the same authors argue that in situ genomic engineering "can dramatically reduce costs and broaden accessibility of these biotechnologies for health and wellness and facilitate equitable access around the world" — the difference between a therapy for a rare disease and one for sickle cell disease across sub-Saharan Africa and India. Without solving the delivery problem, gene therapy will remain limited to rare diseases treatable by liver-targeted or ex vivo approaches.

What’s been tried and why it hasn’t worked

Adeno-associated viruses (AAVs) are the most clinically advanced vectors but face three limitations: pre-existing neutralizing antibodies (NAbs) are common enough to exclude a large share of candidate patients; redosing triggers immune responses; and the roughly 5 kb packaging capacity excludes large genes such as dystrophin, whose coding sequence is several times the AAV limit (Duchenne muscular dystrophy). The largest multi-country seroprevalence survey to date — 502 adults and 50 children across 10 countries, six clinically relevant serotypes — found NAb positivity ranging from 57.8% (AAV9) to 74.9% (AAV1) at 1:1 serum dilution, and from 27.1% (AAV5) to 53.4% (AAV1) at the more stringent 1:4 dilution, with country-level prevalence spanning 36.0% (AAVRh74var in Japan) to 96.0% (AAV1 in South Korea) (Chhabra et al. 2024). Prevalence rose with age, so the patients most likely to need treatment are the ones most likely to be excluded. Lipid nanoparticles (LNPs) — the technology behind mRNA COVID vaccines — work well for liver targeting, but the large majority of an intravenous dose accumulates in the liver regardless of surface modifications. Redirecting LNPs to lung, brain, or muscle tissue has been attempted via ligand conjugation and charge modification, but efficacy in non-liver tissues remains below therapeutic thresholds. CRISPR delivery in vivo faces the additional challenge of delivering both the Cas protein and guide RNA to the same cell simultaneously.

What would unlock progress

Engineered AAV capsid variants (discovered through directed evolution or machine learning-guided design) that evade pre-existing antibodies and exhibit programmable tissue tropism would open the field. For non-viral approaches, understanding the biophysical mechanisms of LNP endosomal escape — image-based single-cell analysis found that siRNA escapes endosomes into the cytosol "at low efficiency (1–2%) and only during a limited window of time" (Gilleron et al. 2013) — could enable rational design of escape-enhancing lipid compositions. A "modular delivery platform" where targeting, immune evasion, and payload release components can be independently optimized and combined would accelerate progress across disease targets.

Entry points for student teams

A student team could use computational approaches (molecular dynamics simulation of lipid-membrane interactions, or machine learning models trained on existing capsid-tropism datasets) to predict which structural modifications to AAV capsids or LNP lipid compositions would improve tissue-specific delivery. Alternatively, teams could design and test novel LNP formulations optimized for non-liver targeting using in vitro cell-type-specific uptake assays with fluorescent reporters. Relevant disciplines: biomedical engineering, chemical engineering, molecular biology, computational biology.